1. Introduction
The island of Oahu has a small land surface area (1536 km2) with a horizontal dimension of about 40 km. The island terrain is dominated by two narrow, nearly parallel mountain ranges, Koolau and Waianae, with peaks of 944 and 1227 m, respectively, and relatively flat terrain over central Oahu (Fig. 1a). Both mountain ridges are oriented in a northwest–southeast direction with tops well below the typical base of the trade wind inversion, which is ~1.9 km over the open ocean (Bingaman 2005; Winning et al. 2017). During the summer months, trade wind persistence, as measured at buoy 51WHO (22.667°N, 157.950°W), is >95% (their Table 2, Hitzl et al. 2014). During the cool season (November–April), mean surface winds upstream of Hawaii become slightly slower and more easterly (Hitzl et al. 2020) and are frequently interrupted by synoptic disturbances, including cold fronts, upper-level troughs, and kona storms (subtropical cyclones) (Blumenstock and Price 1967; Wang and Chen 1998).
(a) A map of the Hawaiian Islands with the three nested domains employed in the simulations with grid sizes of 13.5, 4.5, and 1.5 km, respectively. The blue dot represents the location of the Molokai radar station. (b) The island of Oahu with elevation contours every 200 m. The ×s represent the MesoWest surface weather stations. For the 2003 case study, six stations (red ×s) are used. For the composite study during 2000–15, 15 stations (red and brown ×s) are used. The blue dots represent the Hydronet rain gauge locations. The solid line indicates the locations of the cross-sectional analysis. The gray star and black star represent an upstream point (21.5°N, 157.5°W) and a point over central Oahu (21.5°N, 158.05°W), respectively.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
(a) A map of the Hawaiian Islands with the three nested domains employed in the simulations with grid sizes of 13.5, 4.5, and 1.5 km, respectively. The blue dot represents the location of the Molokai radar station. (b) The island of Oahu with elevation contours every 200 m. The ×s represent the MesoWest surface weather stations. For the 2003 case study, six stations (red ×s) are used. For the composite study during 2000–15, 15 stations (red and brown ×s) are used. The blue dots represent the Hydronet rain gauge locations. The solid line indicates the locations of the cross-sectional analysis. The gray star and black star represent an upstream point (21.5°N, 157.5°W) and a point over central Oahu (21.5°N, 158.05°W), respectively.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
(a) A map of the Hawaiian Islands with the three nested domains employed in the simulations with grid sizes of 13.5, 4.5, and 1.5 km, respectively. The blue dot represents the location of the Molokai radar station. (b) The island of Oahu with elevation contours every 200 m. The ×s represent the MesoWest surface weather stations. For the 2003 case study, six stations (red ×s) are used. For the composite study during 2000–15, 15 stations (red and brown ×s) are used. The blue dots represent the Hydronet rain gauge locations. The solid line indicates the locations of the cross-sectional analysis. The gray star and black star represent an upstream point (21.5°N, 157.5°W) and a point over central Oahu (21.5°N, 158.05°W), respectively.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Heavy rainfall events are major meteorological disasters that occur around the globe. Although occurring in diverse locations, these events share two essential similarities: copious moisture and instability related to lifting (e.g., Ogura et al. 1985; Wang et al. 1985; Kodama and Barnes 1997; Lin et al. 2001; Chiao et al. 2004; Medina et al. 2005; Zhang et al. 2005a; Chen et al. 2018; Tu and Chen 2011; Tu et al. 2019, 2020). One of the greatest challenges facing weather forecasters in the Hawaiian Islands is accurate prediction of heavy rains and flash floods (Schroeder 1977). Except for heavy rainfall caused by tropical storms, these events typically occur during the winter storm season (Haraguchi 1977; Cram and Tatum 1979; Dracup et al. 1991) and are associated with synoptic disturbances (Blumenstock and Price 1967).
Occasionally, during the transition periods between cool and warm seasons (June and October), short-lived localized heavy rainfall events form over central Oahu in the afternoon hours under weak wind (<3 m s−1) conditions determined by wind data at an upstream point from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis (Hartley and Chen 2010). From 2000 to 2015, a total of 13 afternoon heavy rainfall days were reported over central Oahu in the National Centers for Environmental Information (NCEI) storm events database (NCEI 2017). Some of these days are temporally contiguous (e.g., 5 days in 2003, 3 days in 2013, and only 1 day separates the two cases in 2006) (Table 1). It is apparent that large-scale conditions favorable for this type of event are rare during the seasonal transitions, however, if these conditions occur they may last for a few days. Despite their short duration (<4 h), these events may cause small stream and drainage ditch flooding and landslides, which can disrupt traffic and result in property damage. In contrast to heavy orographic precipitation events during the winter storm season (e.g., Schroeder 1977; Tu and Chen 2011), these events occur over central Oahu where terrain is relatively flat and semiarid. With a horizontal island dimension of ~40 km, these localized events are poorly predicted by operational global models with a 0.25° grid size. According to archived forecasts issued by the National Weather Service (NWS) Honolulu office, these events were only predicted on 5 of 13 days (not shown). A better description and understanding of physical processes leading up to initiation of this type of event is needed in order to improve short-range weather forecasting.
A list of the 13 heavy rainfall events, including the date, accumulated rainfall maximum during 0900–1700 HST, and the disaster records in the NCEI storm event database.
Over isolated mountainous tropical and subtropical islands, heavy local showers are frequent with large spatial variations due to terrain and local circulations, including over islands in the Maritime Continent (Qian 2008), Puerto Rico (Jury and Chiao 2013), Taiwan (e.g., Yeh and Chen 1998; Li et al. 1997; Chen et al. 2007; Kerns et al. 2010; Ruppert et al. 2013), Tiwi Islands (Carbone et al. 2000), Dominica (Smith et al. 2009, 2012; Wang and Kirshbaum 2015), and others. Carbone et al. (2000) showed that all mesoscale convective systems over the Tiwi Islands are initiated by a confluence of island-scale sea-breeze fronts or convergence among sea breezes and convective cool pools. Jury and Chiao (2013) simulated leeside thermally driven circulations and afternoon thunderstorms under trade wind conditions over the western slopes of Puerto Rico. Wang and Kirshbaum (2015) simulated thermally forced convection due to elevated heat sources over Dominica (mountaintops ~1.5 km) during a weak wind day (<2 m s−1). Their sensitivity tests confirm that thermal forcing drives convection over the island interior due to localized lifting, which is caused by upslope flow and convergence over the island interior (their Fig. 9).
During summer trade wind weather on Oahu, there are considerable spatial variations in airflow, thermodynamic variables, and rainfall throughout the diurnal cycle despite the island’s relatively small size (Hartley and Chen 2010; Nguyen et al. 2010). Rainfall occurrences in the vicinity of the Koolau Mountain Range have a nocturnal maximum, which is attributed to a combination of nocturnal cloud-top radiative cooling and orographic blocking when the land surface is coldest. During the daytime hours, sea breezes develop along the western Waianae leeside coast, especially under weak trades (Hartley and Chen 2010). Over central Oahu, between the Koolau Mountains and Waianae Mountains, winds are calm in the afternoon with low hourly rainfall frequency (<5%) and very little rainfall overall (<1 mm day−1).
The main goals of this research are to identify favorable large-scale settings, and island-scale features and circulations associated with central Oahu’s heavy rainfall events occurring in the semiarid region during the seasonal transitions (June and October). This paper is organized as follows. The data and methods are given in section 2. Section 3 presents our composite analysis of the synoptic environment favoring this type of storm with a review of large-scale settings present during each event. The model results are first validated using surface and radar observations. Then, the simulated evolution of these events is compared with composite analysis of surface and radar data. A case study is presented in section 4, which provides additional insights into the physical processes involved in the development of this type of storm, including favorable large-scale settings, local conditions, and storm evolution. In section 5, model simulations and sensitivity tests of the case are used to study the physical processes involved in the initiation of this type of event including the mixed layer growth over central Oahu between sunrise and the onset of convection, level at which saturation is first achieved, and effects of daytime land surface heating on the formation of this type of event. The results and findings from this study are summarized in section 6.
2. Data and methods
a. Observations
Large-scale environmental data were sourced from the NCEP Climate Forecast System (CFS). Cases prior to January 2011 used data from CFS Reanalysis (CFSR, Saha et al. 2010), whereas later events used data from CFS version 2 (CFSv2, Saha et al. 2014) operational analysis. Fields were available every 6 h on pressure levels and a 0.5° × 0.5° horizontal grid. Synoptic weather patterns, environmental conditions, and instability were analyzed with these data. They were also used as initial and boundary conditions for our high-resolution numerical simulations.
The Geostationary Operational Environmental Satellite (GOES) infrared (IR) images are used to identify the initiation of convective storms. The NCEI archived Doppler radar observations from the Molokai station (Fig. 1a, blue dot) depict the horizontal distribution of the column-maximum reflectivity from volume scans within range of the radar every 5–7 min. There are 26 NWS Honolulu Forecast Office Hydronet stations with rain gauge data over Oahu from 2000 to 2015 (Fig. 1b, blue dots). The Hydronet rain gauge network reports measurements at 15-min intervals with a resolution of 0.254 mm (0.01 in.). Observations of surface wind speed, surface wind direction, air temperature, and relative humidity (RH) are from the MesoWest cooperative network (Horel et al. 2002) and are quality controlled using MesoWest data quality ratings (Splitt and Horel 1998), with further subjective checks for spatial and temporal consistency during manual analysis. MesoWest observations passed a data quality screening, however, some stations with unrealistic observations were removed manually including stations with unusually high surface air temperature before 1200 Hawaiian standard time (HST = UTC − 10 h) (e.g., KTAH, MKGH, MAPH, MKRH, and SCEH) as compared to the daily maximum temperature observed at PHNL in June and October (~30°C), near saturation during nighttime at HOFH, and low moisture (RH < 10%) at MKGH. MesoWest observations typically represent averages over the reporting interval, which can vary from 5 min to 1 h. The stations with observations within 15 min of and nearest to the verification times are used for model validation (Fig. 1b, crosses).
The storm events database (NCEI 2017) contains the occurrence of storms and other significant weather phenomena documented by the NWS from 1950 to the present. In this study, daytime (e.g., 1000 to 1600 HST) heavy rainfall events during the warm season (May–October) from 2000 to 2015 are selected. During the 16-yr period, a total of 33 days were reported in the database for Oahu. The definition of heavy rainfall follows the NWS criteria from the Storm Data preparation directive (NWS 2016) as an “unusually large amount of rain that led to damage.” The GOES IR images and Molokaʻi radar reflectivities were reviewed to select days in which heavy rain formed over central Oahu. Days in which hurricanes or fronts affected local weather were removed from our sample, and days with wind speeds greater than 3 m s−1 at an upstream point (21.5°N, 157.5°W; gray star in Fig. 1b) based on the CFSR (CFSv2) 1000-hPa level data at 1200 UTC were also excluded. Only 13 days between June and October satisfy our criteria (Table 1). There is no clear correlation between the occurrence of local heavy rainfall and large-scale climate variability (e.g., El Niño–Southern Oscillation).
b. Numerical simulations
The Advanced Research Weather Research and Forecasting Model (WRF-ARW; Skamarock et al. 2008), version 3.6.1, is initialized at 0000 UTC (1400 HST) each day using the NCEP CFSR (CFSv2) data. The total time integration of each simulation is 36 h. Three two-way nested domains are employed (Fig. 1a), with horizontal grid sizes of 13.5, 4.5, and 1.5 km, respectively. There are 38 vertical levels from the surface to the model top around 20 km. The model uses a stretched vertical grid with vertical grid spacing increasing from ~100 m at the surface to ~1500 m near the model top (Chen and Feng 2001; Hitzl et al. 2014).
The physics options include Betts–Miller–Janjić (BMJ) cumulus parameterization scheme (Janjić 1994, 2000), Yonsei University (YSU) planetary boundary layer scheme (Hong et al. 2006), and single-moment 6-class microphysics scheme (Hong and Lim 2006), which uses five prognostic hydrometeor categories (cloud water, cloud ice, rain, snow, and graupel). The cloud droplet condensation number is reduced from the default value of 300 to 50 cm−3 because of the typically clean maritime air found in Hawaii (Hudson 1993). The Rapid Radiative Transfer Model (Mlawer et al. 1997) longwave radiation scheme, Dudhia shortwave radiation scheme (Dudhia 1989), and revised surface layer scheme (Jiménez et al. 2012) were used. The Noah land surface model (Chen and Dudhia 2001) with four soil layers is employed. Further details about the lower boundary conditions over land are given in appendix.
The Control run (CTRL) uses full model physics and high-resolution model terrain with updated vegetation, soil type, initial soil temperature, and soil moisture over the island as given in the appendix. Two sensitivity tests are used to test the impacts of lower boundary conditions over land on the initiation of the central Oahu storm. The NoOrog run has the same updated lower boundary conditions over the island as the CTRL run but with island topography reduced to sea level to assess the effects of orographic blocking and lifting. The NoFlux run is similar to the CTRL run, but the surface heat and moisture fluxes are turned off to test the effects of thermal forcing from the land surface.
3. A composite study
In this section, we examine the general characteristics of our 13 heavy rain days using composite analysis. We also assess the accuracy of our numerical simulations through a quantitative comparison with both surface station data and radar reflectivities.
a. Synoptic conditions
From the review of large-scale settings, the precursors of this type of storm are an upper-level short-wave trough coming down from the northwest with a blocking high over the mid-Pacific in midlatitudes, a weak surface wind shear line, a moisture tongue with total precipitable water (TPW) > 40 mm over the Hawaiian region, and weak low-level winds. To depict the large-scale settings of these rare storms, a synoptic analysis of a 13-day composite is presented (Fig. 2).
(a) The NCEP CFSR (CFSv2) 300-hPa geopotential height (gpm, contours) and wind barbs (pennant: 50 m s−1; full barb: 10 m s−1; half barb: 5 m s−1; open circle: calm) at 1400 HST for the 13-day composite. The red cross represents the location of Oahu and the dashed line represents a trough. (b) The NCEP CFSR total precipitable water (mm, color shades), surface wind barbs (pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm), and mean sea level pressure (hPa, contours) at 1400 HST for the 13-day composite. The black cross represents the location of Oahu. (c) Skew T–logp diagram from NCEP CFSR (CFSv2) at 21.5°N, 157.5°W at 1400 HST for the 13-day composite. Pennants, full barbs, half barbs, and open circles represent 5 m s−1, 1 m s−1, 0.5 m s−1, and calm, respectively. CAPE is calculated from the surface using the virtual temperature.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
(a) The NCEP CFSR (CFSv2) 300-hPa geopotential height (gpm, contours) and wind barbs (pennant: 50 m s−1; full barb: 10 m s−1; half barb: 5 m s−1; open circle: calm) at 1400 HST for the 13-day composite. The red cross represents the location of Oahu and the dashed line represents a trough. (b) The NCEP CFSR total precipitable water (mm, color shades), surface wind barbs (pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm), and mean sea level pressure (hPa, contours) at 1400 HST for the 13-day composite. The black cross represents the location of Oahu. (c) Skew T–logp diagram from NCEP CFSR (CFSv2) at 21.5°N, 157.5°W at 1400 HST for the 13-day composite. Pennants, full barbs, half barbs, and open circles represent 5 m s−1, 1 m s−1, 0.5 m s−1, and calm, respectively. CAPE is calculated from the surface using the virtual temperature.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
(a) The NCEP CFSR (CFSv2) 300-hPa geopotential height (gpm, contours) and wind barbs (pennant: 50 m s−1; full barb: 10 m s−1; half barb: 5 m s−1; open circle: calm) at 1400 HST for the 13-day composite. The red cross represents the location of Oahu and the dashed line represents a trough. (b) The NCEP CFSR total precipitable water (mm, color shades), surface wind barbs (pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm), and mean sea level pressure (hPa, contours) at 1400 HST for the 13-day composite. The black cross represents the location of Oahu. (c) Skew T–logp diagram from NCEP CFSR (CFSv2) at 21.5°N, 157.5°W at 1400 HST for the 13-day composite. Pennants, full barbs, half barbs, and open circles represent 5 m s−1, 1 m s−1, 0.5 m s−1, and calm, respectively. CAPE is calculated from the surface using the virtual temperature.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
At the 300-hPa level, the composite synoptic geopotential height exhibits an omega blocking pattern (Fig. 2a). A shortwave trough was located over the Hawaiian Islands. As a result of the blocking pattern, the conditions necessary for this type of storm can last 3–5 days. At the surface, a shear line and the associated axis of high TPW (>40 mm) were located over the Hawaiian Islands (Fig. 2b). The composite skew T–logp plot for the designated upstream point (Fig. 2c) shows that the environmental conditions are characterized by the following: 1) a conditionally unstable atmosphere (below 500 hPa) with the equilibrium layer (EL) at 200 hPa, 2) absence of a trade wind inversion, and 3) a deep moist layer with relatively high (>40 mm) TPW, and convective available potential energy (CAPE) of ~382 J kg−1 (Fig. 2c). CAPE is calculated using the virtual temperature and mixing ratio of an air parcel originating from the surface. The mean winds at low levels are weak southeasterlies (<3 m s−1) (Fig. 2c).
b. Island-scale analysis
From the 13-day composite analysis, weak onshore flows (~1 m s−1) are observed over Oahu’s coasts at 0900 HST with calm winds over central Oahu (Fig. 3a). At 1000 HST, the onshore flows over the southern coast increase to 3–4 m s−1 (Fig. 3b). Prior to the onset of convection (~2–3 h) over central Oahu, the onshore flows from the northwestern and southeastern coasts converge over central Oahu (Figs. 3c,d). At 1600 HST, weakened onshore flows over the northern and southern coasts (~1.5 m s−1, Fig. 3i) are still observed.
Mean observed radar echoes (dBZ, color shades) and mean observed surface wind (wind barb, pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm) for the 13-day composite at (a) 0900, (b) 1000, (c) 1100, (d) 1200, (e) 1300, (f) 1400, (g) 1500, (h) 1600, and (i) 1700 HST. The black box indicates the region used for the area average.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Mean observed radar echoes (dBZ, color shades) and mean observed surface wind (wind barb, pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm) for the 13-day composite at (a) 0900, (b) 1000, (c) 1100, (d) 1200, (e) 1300, (f) 1400, (g) 1500, (h) 1600, and (i) 1700 HST. The black box indicates the region used for the area average.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Mean observed radar echoes (dBZ, color shades) and mean observed surface wind (wind barb, pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm) for the 13-day composite at (a) 0900, (b) 1000, (c) 1100, (d) 1200, (e) 1300, (f) 1400, (g) 1500, (h) 1600, and (i) 1700 HST. The black box indicates the region used for the area average.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Reflectivities at each hour during 1000–1700 HST are used to make a 13-day composite. During 0900–1100 HST, scattered radar echoes of ~10 dBZ first occur over the two mountain ranges (Figs. 3a–c). Heavy showers (~30 dBZ) over central Oahu are first observed around 1200 HST (Fig. 3d). At 1400 HST, the composite radar echoes reach 40 dBZ over central Oahu with onshore flows along all coasts (Fig. 3f). The radar echoes over the island interior weaken during 1500–1600 HST and dissipate by 1700 HST (Fig. 3i).
c. Model verification
Model verification during 0900–1100 is conducted in order to test model performance prior to convection initiation. Error statistics for hourly 2-m air temperature, 2-m RH, and 10-m winds for these 13 days are given in Table 2 and Table 3, respectively. The simulated temperature is adjusted with a lapse rate of 6.5 K km−1 to account for the height difference between the model terrain and the station height. The maximum difference is 167.6 m at KTAH (Table 2). The magnitudes of both MAE (−0.6°C) and RMSE (2.0°C) for all stations are comparable to Nguyen et al. (2010) using the MM5 model. The RH error statistics at 13 stations show that the model underestimates moisture at seven stations (bias from −0.3% to −17.6%) with RMSE ranging from 4.2% to 21.4%. The largest errors occur at OFRH with cool mean surface temperature (~24°C) and high mean surface RH (90%) at 1100 HST (not shown). Station OFRH is located in a high moisture and cloudy environment, resulting from the drifting of orographic showers from the ridge tops of the Koolau Mountain Range to the western upper leeside slopes (Nguyen et al. 2010), which is not well simulated by the model.
Error statistics for surface temperature and RH reported during 0900–1100 HST for the 13 heavy rainfall days. The error statistics are computed every hour.
Error statistics for surface wind reported during 0900–1100 HST for the 13 heavy rainfall days. The error statistics are computed every hour.
The RAWS stations measure surface wind at 6 m above ground, whereas the winds from the NWS stations are measured at 10 m. The surface wind from the RAWS station is adjusted to 10 m with a correction factor of 1.086 (Bradshaw et al. 2003; Hartley and Chen 2010). The largest RMSE occurred at stations located near the island corners (KFWH, PHJR, and KKRH), where the tip jet effect (Reeve and Kolstad 2011; Hitzl et al. 2014) is most significant (Table 3). Other stations have RMSE values in the range of 0.9 to 2.0 m s−1. Wind errors at stations over central Oahu and the southern coast are relatively small (RMSE < 1.8 m s−1). Overall, the simulated surface winds are in reasonable agreement with observations.
To verify the timing and location of simulated convection, a boxed region over central Oahu is selected (Fig. 3f). For 12 October and 14 October 2013, there were no rainfall observations over central Oahu. Thus, area averages of the composite observed radar reflectivities and the composite simulated radar reflectivities are calculated within the box. Figure 4 shows the scatterplot of area-averaged observed radar reflectivities and area-averaged simulated radar reflectivities at various times of the day. Overall, the model slightly overestimates reflectivities during 1100–1700 HST (Fig. 4).
Scatterplot of area-average observed radar echoes and area-average simulated radar echoes. The dotted line is a line of best fit.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Scatterplot of area-average observed radar echoes and area-average simulated radar echoes. The dotted line is a line of best fit.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Scatterplot of area-average observed radar echoes and area-average simulated radar echoes. The dotted line is a line of best fit.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
d. Numerical simulations
Hourly simulated composite surface winds and radar echoes during 0900–1700 HST are presented in Fig. 5. Figure 5a shows that the prevailing winds over the coastal waters are weak southeasterlies. At 1000 HST, onshore flows of ~2–3 m s−1 are simulated over Oahu’s coasts (Fig. 5b). At 1100 HST, the speed of the onshore flows over the northern and southern coasts increases to 4–5 m s−1 (Fig. 5c). Scattered radar echoes of ~30 dBZ are simulated over the two mountain ranges and central Oahu (Fig. 5c). During 1200–1300 HST, widespread radar echoes are simulated over the island interior (Figs. 5d,e). At 1400 HST, radar echoes > 40 dBZ are simulated over central Oahu (Fig. 5f). These echoes last for two more hours before dissipating around 1700 HST (Figs. 5g–i).
As in Fig. 3, but for the simulated reflectivity and surface wind.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 3, but for the simulated reflectivity and surface wind.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 3, but for the simulated reflectivity and surface wind.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Both observations (Fig. 3) and simulations (Fig. 5) show that these heavy rainfall cases begin with light showers over the Koolau and Waianae Mountains at 1000 HST with onshore flows from the coasts. Subsequently, the onshore flows from the northwestern and southeastern coasts converge over central Oahu, initiating the central Oahu storm with a northeast–southwest elongated feature at 1100 HST. The simulated radar reflectivities at the initial stage are consistent with the observed patterns, albeit slightly overestimated. Observations show that these short-lived heavy rainfall events end by 1600 HST. However, the simulated heavy rainfall lasts about an hour longer (~1700 HST). Furthermore, the most intense radar reflectivities during 1400–1500 HST are more confined to central Oahu in observations than in simulations.
4. A case study of the central Oahu storm (8 June 2003)
Although there are many similarities between the days shown in the composite of section 3, large-scale settings, local conditions, and storm evolutions vary somewhat case by case. In this section, the case that occurred on 8 June 2003 is selected for detailed study. The first echo was detected ~1100 HST and the storm dissipated ~1500 HST. Observational analysis and model simulations start from 0700 HST to investigate transition of island-scale diurnal airflows after sunrise and preconvection conditions.
a. Synoptic conditions
On 2 June 2003, a Rex blocking pattern (Rex 1950) is evident at the 300-hPa level, which evolves into an omega-blocking pattern on 6 June (not shown). On 8 June, the upper-level trough (T) associated with the blocking pattern is in the vicinity of the Hawaiian Islands (Fig. 6a). Furthermore, the northerly flow behind the upper-level trough (T) advects cold air over the state (Fig. 6a). At 0200 HST, the horizontal distribution of TPW has a northeast–southwest maximum axis (>43 mm) ahead of the trough that reaches Oahu from the northwest around 1400 HST (Fig. 6b).
As in Fig. 2, but at 1400 HST 8 Jun 2003.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 2, but at 1400 HST 8 Jun 2003.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 2, but at 1400 HST 8 Jun 2003.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
The skew T–logp diagram of an upstream point (the black star in Fig. 1b) from CFSR at 1400 HST 8 June indicates weak southeasterly winds (~3 m s−1) at low levels without the presence of a trade wind inversion (Fig. 6c). Winds are easterlies/northeasterlies (~5 m s−1) between 900 and 700 hPa, becoming northwesterly/northerly (~15 m s−1) above 700 hPa. The temperature profile is approximately pseudoadiabatic above 500 hPa with a positive (907 J kg−1) CAPE much greater than the composite value. However, the storm environment is consistent with common features found in the composite analysis, including a blocking pattern in midlatitudes, a northeast–southwest moist tongue (TPW > 40 mm) from low latitudes ahead of an upper-level trough, and absence of a trade wind inversion. The EL is at 200 hPa, indicating the cloud top may reach 12 km.
b. Island-scale analysis
At 0700 HST, offshore flows were observed at coastal stations (Fig. 7a). Over the island interior, mountain winds were evident on the eastern slopes of the Waianae Mountains and the western slopes of the Koolau Mountains. At SCBH, which is on the eastern foothills of the Waianae Mountains (Fig. 1b), the southwesterly katabatic flow becomes northerly anabatic flow after 0900 HST as a result of daytime heating (Fig. 7b). At PHNL, the offshore flow at 0800 HST (not shown) is replaced by a sea breeze after 0900 HST (Fig. 7b). At all other Oahu coastal stations, the offshore flows (Fig. 7a) shift to onshore flows well before 1100 HST (Fig. 7d).
Observed radar reflectivity (dBZ, color shades) and surface wind (wind barbs, pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, (f) 1300, (g) 1400, (h) 1500, and (i) 1600 HST 8 Jun.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Observed radar reflectivity (dBZ, color shades) and surface wind (wind barbs, pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, (f) 1300, (g) 1400, (h) 1500, and (i) 1600 HST 8 Jun.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Observed radar reflectivity (dBZ, color shades) and surface wind (wind barbs, pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, (f) 1300, (g) 1400, (h) 1500, and (i) 1600 HST 8 Jun.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
At 1100 HST, weak, isolated radar echoes were detected on the northwestern leeside slopes of the Koolau Mountains (Fig. 7d). At 1200 HST, an arc-shaped radar echo was observed over central Oahu (Fig. 7e). Radar echoes also developed over the northwestern foothills of the Koolau Mountains at this time (Fig. 7e). At 1300 HST, radar echoes covered most of central Oahu (Fig. 7f) and then moved southwestward (Figs. 7g–i). At 1500 HST, the strongest radar echoes were off the Waianae coast (Fig. 7h).
Based on GOES IR images, a cloudy area associated with the weak shear line, with cloud-top brightness temperatures < 280 K (cloud top ~3 km according to Fig. 9b), passed Oahu from the northwest to the southeast between 0700 and 1300 HST (Figs. 8a–f). At 1200 HST, convective clouds with brightness temperatures of ~274 K and cloud top height of ~4 km (Fig. 9b) were observed over central Oahu (Fig. 8e). These clouds developed into deep convection with brightness temperatures of ~256 K at 1300 HST (Fig. 8f) and cloud tops of 6 km (Fig. 9b). At 1400 HST, clouds with cloud-top temperatures < 220 K covered most of central Oahu (Fig. 8g). Based on the simulated sounding over central Oahu (Fig. 9b), the temperature of the EL (~11.5 km) is ~215 K, which is consistent with GOES IR images (Fig. 8g). The highest echoes (~10 km) occurred over the Waianae Mountains as the storm moved southwestward (not shown). During the later stages, the cirrus clouds associated with the convection moved southeastward (Figs. 8h,i) with the upper-level northwesterly winds (Fig. 9b).
IR brightness temperature (K, color shades) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, (f) 1300, (g) 1400, (h) 1500, and (i) 1600 HST 8 Jun.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
IR brightness temperature (K, color shades) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, (f) 1300, (g) 1400, (h) 1500, and (i) 1600 HST 8 Jun.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
IR brightness temperature (K, color shades) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, (f) 1300, (g) 1400, (h) 1500, and (i) 1600 HST 8 Jun.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Skew T–logp diagram from WRF simulations at central Oahu (21.5°N, 158.05°W) at (a) 0700 and (b) 1100 HST 8 Jun. Pennants, full barbs, half barbs, and open circles represent 5 m s−1, 1 m s−1, 0.5 m s−1, and calm, respectively. The red dashed line represents the temperature profile of a rising air parcel.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Skew T–logp diagram from WRF simulations at central Oahu (21.5°N, 158.05°W) at (a) 0700 and (b) 1100 HST 8 Jun. Pennants, full barbs, half barbs, and open circles represent 5 m s−1, 1 m s−1, 0.5 m s−1, and calm, respectively. The red dashed line represents the temperature profile of a rising air parcel.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Skew T–logp diagram from WRF simulations at central Oahu (21.5°N, 158.05°W) at (a) 0700 and (b) 1100 HST 8 Jun. Pennants, full barbs, half barbs, and open circles represent 5 m s−1, 1 m s−1, 0.5 m s−1, and calm, respectively. The red dashed line represents the temperature profile of a rising air parcel.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
5. Numerical simulations of the 8 June 2003 case
Model simulations are used as a diagnostic tool to study the physical processes involved in the initiation and maintenance of this type of storm, including the mixed layer growth over central Oahu between sunrise and the onset of convection, local convergence over central Oahu, verification of simulated rainfall, and model sensitivity tests.
a. Prestorm conditions and the evolution of the mixed layer
The simulated skew T–logp chart over central Oahu (Fig. 1b, black star) at 0700 HST on 8 June indicates weak low-level southeasterly winds (<2 m s−1) with no CAPE (Fig. 9a). At 1100 HST, the environment is saturated between 860 and 800 hPa with CAPE of ~555 J kg−1 (Fig. 9b). The TPW value (~40 mm) is significantly greater than the climatological value for June (31–32 mm). Weak northerly winds (<2 m s−1) are present below 860 hPa (Fig. 9b), becoming east-northeasterly between 840 and 700 hPa, turning to north-northwesterlies above 700 hPa. The vertical wind profile and the evolution of radar echoes suggest that precipitation is advected by the east-northeasterly flows above 840 hPa to the southwestern coast (Figs. 7g–i).
The evolution of the mixed layer over central Oahu after 0700 HST is depicted by the simulated vertical profiles of thermodynamic variables (Fig. 10). By 0800 HST, the nocturnal lower-level stability is eliminated (Figs. 10a,b). From 0900 to 1100 HST, low-level potential temperature (θ) continues to increase with deepening of the mixed layer (Fig. 10a). At 1100 HST, the mixed layer top is ~1.5 km above sea level with homogenous θ (~300 K) below that level (Fig. 10a). From 0800 to 1100 HST, the onshore flow brings in maritime air with higher mixing ratio as the mixed layer deepens (Fig. 10c). Convective inhibition (CIN) is zero over central Oahu at 0700 HST, evolving to a CAPE value of 555 J kg−1 at 1100 HST. In the meantime, the LCL and LFC are raised to 1.1 km during the deepening of the mixed layer (Fig. 10e). Saturation between 1.4 and 2 km is simulated (Fig. 10d).
Vertical profiles at central Oahu (21.5°N, 158.05°W) of (a) potential temperature (K), (b) equivalent potential temperature (K), (c) mixing ratio (g kg−1), and (d) relative humidity (%) from 0700 to 1100 HST 8 Jun. (e) Time series of the LCL (solid line) and CAPE (dashed line) at central Oahu (21.5°N, 158.05°W) during 0700–1100 HST 8 Jun. CAPE is calculated from the surface using the virtual temperature.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Vertical profiles at central Oahu (21.5°N, 158.05°W) of (a) potential temperature (K), (b) equivalent potential temperature (K), (c) mixing ratio (g kg−1), and (d) relative humidity (%) from 0700 to 1100 HST 8 Jun. (e) Time series of the LCL (solid line) and CAPE (dashed line) at central Oahu (21.5°N, 158.05°W) during 0700–1100 HST 8 Jun. CAPE is calculated from the surface using the virtual temperature.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Vertical profiles at central Oahu (21.5°N, 158.05°W) of (a) potential temperature (K), (b) equivalent potential temperature (K), (c) mixing ratio (g kg−1), and (d) relative humidity (%) from 0700 to 1100 HST 8 Jun. (e) Time series of the LCL (solid line) and CAPE (dashed line) at central Oahu (21.5°N, 158.05°W) during 0700–1100 HST 8 Jun. CAPE is calculated from the surface using the virtual temperature.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
In summary, the vertical profiles of simulated thermodynamic variables over central Oahu show that, after sunrise, the mixed layer deepens with increasing temperature and moisture due to surface heat and moisture fluxes. Then, land–sea thermal contrast drives sea-breeze/onshore flows, which bring in maritime air over central Oahu (Figs. 7a–c). In the meantime, the LCL (~1.1 km) (Fig. 10e) is just below the top of the mixed layer (~1.4 km) (Fig. 9b) with saturation between 1.4 and 2 km (Fig. 10d) due to rising motion.
b. The evolution of simulated surface winds and vertical motions
At the surface, vertical motion due to orographic lifting is defined as
where v is the simulated surface horizontal wind (m s−1) and h is the model terrain height (m). In this section, the horizontal distribution of vertical motion at the surface due to orographic lifting is compared with the horizontal distribution of vertical motion at the 900-hPa level, which is above the ridgetops.
At 1000 HST, northwesterly sea breezes from the north shore and southeasterly sea breezes from the southern coast converge over central Oahu with vertical motions (~1 m s−1) at the 900-hPa level (Fig. 11a). At 1100 HST, the vertical motion there exceeds 1 m s−1 (Fig. 11b). The first radar echo over central Oahu was observed at 1200 HST. Thus, the time lag between surface convergence and the first echo is about 2–3 h. At 1100 HST, rising motion at the 900-hPa level is also simulated over the western leeside slopes of the Koolau Mountains, where the combined easterly/upslope flow moves over the ridge tops and converges with the westerly upslope winds from central Oahu (Fig. 11b), similar to the convergence simulated by Wang and Kirshbaum (2015) over Dominica. Over the Waianae Mountains, rising motion is simulated near the ridge tops where the onshore/upslope flow from the western coast converges with the valley winds from the eastern slopes. At the slope surface, rising motion (wsfc) occurs mainly immediately above the eastern slopes of the Koolau Mountains and the western slopes of the Waianae Mountains due to orographic lifting of the onshore/upslope flows. However, vertical motions at the 900-hPa level due to orographic lifting of the onshore/upslope flows are insignificant (Fig. 11).
Simulated vertical motion at 900 hPa (m s−1, color shade), simulated surface vertical motion due to orographic lifting (wsfc) (m s−1, black contours), simulated surface wind (m s−1, wind barb), and model terrain height (gray contour, 100-m interval) at (a) 1000 and (b) 1100 HST 8 Jun. Pennants, full barbs, half barbs, and open circles represent 5 m s−1, 1 m s−1, 0.5 m s−1, and calm, respectively.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Simulated vertical motion at 900 hPa (m s−1, color shade), simulated surface vertical motion due to orographic lifting (wsfc) (m s−1, black contours), simulated surface wind (m s−1, wind barb), and model terrain height (gray contour, 100-m interval) at (a) 1000 and (b) 1100 HST 8 Jun. Pennants, full barbs, half barbs, and open circles represent 5 m s−1, 1 m s−1, 0.5 m s−1, and calm, respectively.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Simulated vertical motion at 900 hPa (m s−1, color shade), simulated surface vertical motion due to orographic lifting (wsfc) (m s−1, black contours), simulated surface wind (m s−1, wind barb), and model terrain height (gray contour, 100-m interval) at (a) 1000 and (b) 1100 HST 8 Jun. Pennants, full barbs, half barbs, and open circles represent 5 m s−1, 1 m s−1, 0.5 m s−1, and calm, respectively.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Figure 12 shows a northwest–southeast vertical cross section (Fig. 1b) of equivalent potential temperature (θe) and winds. At 0700 HST, weak offshore flows with relatively low θe are simulated along the northern coast (Fig. 12a). At 0900 HST, northerly onshore flows from the northern coast and southerly onshore flows from the southern coast develop (Fig. 12b). Sea breezes from the northern and the southern coasts reach central Oahu at 1000 HST (Fig. 12c). The air advected from the south shore has slightly higher θe compared to the air coming from the north shore (Figs. 12a–c). At 1100 HST, the sea breezes with high θe (338–340 K) converge below the 1-km level over central Oahu with divergence above (1–2.5 km) (Fig. 12d). The rising motion (>1 m s−1) transports the low-level, high θe air above the 1-km level (Fig. 12d). The convergence over central Oahu provides the localized lifting for the release of instability and the development of deep convection (Figs. 7e and 13e). There is evidence of convergence between the sea breezes and convective cool pool in the early afternoon (Figs. 12e,f).
Vertical cross section along the northwest–southeast line in Fig. 1b showing equivalent potential temperature (K, color shades), horizontal wind speed along the cross section [gray contours with 1 m s−1 intervals; zero contours are omitted; positive (solid contours) pointing in the northwestern direction], vertical motion (black contours with 0.5 m s−1, 1 m s−1, and 2 m s−1 intervals) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, and (f) 1300 HST 8 Jun. The origin of the x axis represents the southernmost point of the northwest–southeast line in Fig. 1b.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Vertical cross section along the northwest–southeast line in Fig. 1b showing equivalent potential temperature (K, color shades), horizontal wind speed along the cross section [gray contours with 1 m s−1 intervals; zero contours are omitted; positive (solid contours) pointing in the northwestern direction], vertical motion (black contours with 0.5 m s−1, 1 m s−1, and 2 m s−1 intervals) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, and (f) 1300 HST 8 Jun. The origin of the x axis represents the southernmost point of the northwest–southeast line in Fig. 1b.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Vertical cross section along the northwest–southeast line in Fig. 1b showing equivalent potential temperature (K, color shades), horizontal wind speed along the cross section [gray contours with 1 m s−1 intervals; zero contours are omitted; positive (solid contours) pointing in the northwestern direction], vertical motion (black contours with 0.5 m s−1, 1 m s−1, and 2 m s−1 intervals) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, and (f) 1300 HST 8 Jun. The origin of the x axis represents the southernmost point of the northwest–southeast line in Fig. 1b.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
c. The evolution of simulated radar reflectivities
Before noon, the horizontal distributions of observed and simulated radar reflectivities reveal scattered echoes along the two mountain ranges. More scattered showers are simulated along the western slope of the Koolau Mountain Range than are observed during this period (Figs. 13b–d). At noon, the observed and simulated radar reflectivities both show arc-shaped echoes over central Oahu with maximum reflectivities > 40 dBZ (Figs. 13e and 7e). At 1400 HST, the simulated radar echoes drift southwestward (Fig. 13g) due to the northeasterlies between 870 and 700 hPa (Fig. 9b) with simulated heavy rainfall over the southwestern coast. At 1500 HST, the observed convection persists and has moved southwestward to the open ocean (Fig. 7h), whereas the simulated storm rapidly dissipates (Fig. 13h). Simulated local showers continue to develop over the western slopes of the Koolau Mountain Range (Fig. 13i), which are not observed.
Simulated radar reflectivity (dBZ, color shades) and model terrain height (gray contours, 100-m intervals) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, (f) 1300, (g) 1400, (h) 1500, and (i) 1600 HST 8 Jun. Wind barbs represent the simulated surface wind (pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm).
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Simulated radar reflectivity (dBZ, color shades) and model terrain height (gray contours, 100-m intervals) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, (f) 1300, (g) 1400, (h) 1500, and (i) 1600 HST 8 Jun. Wind barbs represent the simulated surface wind (pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm).
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Simulated radar reflectivity (dBZ, color shades) and model terrain height (gray contours, 100-m intervals) at (a) 0700, (b) 0900, (c) 1000, (d) 1100, (e) 1200, (f) 1300, (g) 1400, (h) 1500, and (i) 1600 HST 8 Jun. Wind barbs represent the simulated surface wind (pennant: 5 m s−1; full barb: 1 m s−1; half barb: 0.5 m s−1; open circle: calm).
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
d. Simulated rainfall in comparison with observations
From 1100 to 1600 HST 8 June, a large amount of precipitation was recorded by the Hydronet stations in central Oahu, with rain rates of ~48 mm h−1 (not shown). Most of the rainfall was recorded during 1400–1600 HST, with a maximum rainfall accumulation of ~58 mm (Fig. 14). The simulated precipitation over central Oahu is less (~30 mm) than the recorded rain gauge measurements (Fig. 14). The observed local rainfall maximum over the Waianae Mountains (~50 mm) is also simulated.
Accumulation of simulated rainfall (mm, color shades) and accumulation of rain gauge observations (mm, colored dots with label) from 1100 to 1600 HST 8 Jun. Rain gauges with no rainfall accumulation (0.0 mm) are shown as black dots.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Accumulation of simulated rainfall (mm, color shades) and accumulation of rain gauge observations (mm, colored dots with label) from 1100 to 1600 HST 8 Jun. Rain gauges with no rainfall accumulation (0.0 mm) are shown as black dots.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Accumulation of simulated rainfall (mm, color shades) and accumulation of rain gauge observations (mm, colored dots with label) from 1100 to 1600 HST 8 Jun. Rain gauges with no rainfall accumulation (0.0 mm) are shown as black dots.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
e. Model sensitivity tests
In the Dominica case study by Wang and Kirshbaum (2015), the convection occurs over the slope where the upslope flow from the western slope moves over the ridgetop and converges with the upslope flow from the other side of the mountain. For our case, the strongest vertical motion occurs over central Oahu with a flat terrain without orographic lifting. Note that the onshore/sea-breeze flows from the northwestern and southeastern shores are about 4–6 m s−1 (Fig. 7), which are considerably stronger than the upslope flows along both sides of the mountains (1–3 m s−1) bordering the central Oahu plain. Onshore flows coming from the northwestern and southeastern shores and penetrating the semiarid flat terrain are linked to the deepening of the mixed layer (~1.5 km) over central Oahu (Fig. 10). As a result, the most intense convective showers are initiated in this convergence zone over central Oahu (Figs. 7 and 8).
Figure 15 shows the evolution of simulated radar echoes from the NoOrog run on 8 June. Thermally driven sea breezes begin developing at 1000 HST with rising motion along the leading edge of the sea breezes (Figs. 15c and 16a). Scattered radar echoes are simulated over Central and Western Oahu at 1100 HST (Fig. 15d). The vertical motion shows that low-level lifting is primarily caused by low-level convergence of sea-breeze circulations (Figs. 16b,c). At 1300 HST, an arc-shaped radar echo is simulated over central Oahu (Fig. 15f) in response to low-level lifting (Figs. 16b,c). Carried by midlevel easterlies (Fig. 9), the simulated convection moves westward, as seen in the observed radar echoes (Figs. 15g–i and 7g–i). Convective feedbacks on low-level airflow are evident after 1200 HST (Figs. 16c,d). For the NoOrog run, the timing of convective development is close to what was observed. However, the location of convective initiation is slightly westward as the onshore flow from the east, in the absence of orographic blocking by the Koolau Mountains, is stronger. The results of the NoOrog run indicate that orographic lifting is not essential to the development of this type of heavy rainfall event. In contrast, Nguyen et al. (2010) showed that without terrain, trade wind showers fail to develop over both the Koolau and Waianae Mountains. Their results attest to the importance of orographic lifting in the production of summer trade wind rainfall over Oahu. Nevertheless, strong trade wind days do not always produce more trade wind rainfall than weak trade wind days. Hartley and Chen (2010) showed that most of the orographic showers on the windward side of Oahu are related to preexisting trade wind cumuli or showers that drift inland and are enhanced by the terrain.
As in Fig. 13, but for the no orographic (NoOrog) run.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 13, but for the no orographic (NoOrog) run.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 13, but for the no orographic (NoOrog) run.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 12, but for the NoOrog run at (a) 1000, (b) 1100, (c) 1200, and (d) 1300 HST.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 12, but for the NoOrog run at (a) 1000, (b) 1100, (c) 1200, and (d) 1300 HST.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 12, but for the NoOrog run at (a) 1000, (b) 1100, (c) 1200, and (d) 1300 HST.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
In the NoFlux run, the simulated winds over the windward slopes and central Oahu are weak and variable during 0700–1000 HST (Figs. 17a–c). Sea-breeze/onshore flows are absent in the NoFlux run. Vertical motion over central Oahu is not simulated (not shown). The NoFlux run does not produce convective cells over Oahu (Fig. 17). Without surface heat and moisture fluxes, the lifting mechanism provided by the onshore/sea-breeze convergence over central Oahu is absent. These results attest that onshore flows driven by land–sea thermal contrast are crucial to the increase in instability and the generation of low-level convergence, leading to the initiation of convection over central Oahu. The same sensitivity tests (NoOrog and NoFlux) have been performed for all 13 days with similar results (not shown).
As in Fig. 13, but for the no surface flux (NoFlux) run.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 13, but for the no surface flux (NoFlux) run.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
As in Fig. 13, but for the no surface flux (NoFlux) run.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
6. Discussion and summary
The central Oahu storm is initiated by thermal forcing from the land surface. There are three types of lifting associated with this type of storm: 1) lifting by the upslope flow on the eastern slope of the Koolau Mountains and on the western slope of the Waianae Mountains; 2) convergence on the western slope of the Koolau Mountains and on the eastern slope of the Waianae Mountains; 3) convergence over central Oahu. For orographic lifting of the upslope flows (Type 1), the advection time scale for the maritime air to reach the ridgetops from the coasts is too short (10 min or less) to produce orographic showers (Hartley and Chen 2010). In fact, at 1100 HST, the vertical motion due to orographic lifting above the slope surface by these upslope flows is insignificant. For the convergence zones where the upslope flows (1–2 m s−1) from the coasts move over the ridgetops and converge with the upslope flows (~1 m s−1) from the island interior, scattered showers are generated there. This type of thermally driven convection is similar to the case reported by Wang and Kirshbaum (2015). The mixed layer over the slope surface is rather shallow (<400 m above the slope surface) (not shown) as the daytime heating over the tropical rain forest is reduced by the cold advection associated with the upslope flows (de Wekker and Kossmann 2015). The most intense storm occurs over central Oahu where the sea breezes (3–5 m s−1) from the northwestern coast converge with the sea breezes (4–6 m s−1) from the southeastern shore. Air over the semiarid central Oahu is bound by two narrow mountain ranges with continuous heating after sunrise. For the 8 June 2003 case, the mixed layer grows to 1.5 km above sea level, which is above both the LCL and LFC (~1.1 km), with saturation from 1.4 km to above 2 km at 1100 HST. With persistent convergence over central Oahu, deep convection finally develops after 1200 HST (Fig. 18).
A schematic diagram showing the island-induced airflow involved in the development of the short-lived afternoon heavy rainfall events over central Oahu under weak wind conditions (<3 m s−1) and favorable large-scale settings. The red circle shows the area with low-level convergence.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
A schematic diagram showing the island-induced airflow involved in the development of the short-lived afternoon heavy rainfall events over central Oahu under weak wind conditions (<3 m s−1) and favorable large-scale settings. The red circle shows the area with low-level convergence.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
A schematic diagram showing the island-induced airflow involved in the development of the short-lived afternoon heavy rainfall events over central Oahu under weak wind conditions (<3 m s−1) and favorable large-scale settings. The red circle shows the area with low-level convergence.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
A diurnal heating cycle is present over Oahu throughout the year, especially during the summer months. However, central Oahu storms occur mainly during the seasonal transitions (June and October) and are rather uncommon (totaling 13 days from 2000 to 2015). For instance, Hartley and Chen (2010) found no convection over central Oahu during the summer months under weak wind conditions. Thus, sea-breeze convergence alone is inadequate to initiate central Oahu storms. Favorable large-scale conditions are needed, which are most likely to be met during the seasonal transitions, including: deep moist layer; blocking pattern in the midlatitudes with a northeast–southwest moist tongue with relatively high TPW (>40 mm) ahead of an upper-level trough; absence of a trade wind inversion; and weak low-level winds.
Based on our study, determining the locations and timings of deep convection over mountainous tropical islands requires careful analyses of the synoptic-scale environment as well as the island-scale flow response under the given large-scale settings. Local forcing may be rather complex depending on island size, height, and terrain features, which vary among different tropical or subtropical islands.
The central Oahu storm is difficult to forecast using operational global models because of their relatively coarse model resolution. In this study, we demonstrate the value of island-scale high-resolution models in predicting this type of storm. However, if the large-scale environment is poorly predicted by the global models, it is unlikely that the high-resolution models will correctly forecast the island-scale response. Yet, based on the synoptic precursors delineated in this study and using observations from early morning, it may be possible to nowcast occurrences of the central Oahu storm in the early afternoon.
A well-validated high-resolution model is a valuable tool for studying the island-scale flow response as well as island effects on rainfall, weather, and climate over different island communities under various large-scale settings. Research regarding the effects of different summer trade wind conditions (Yang et al. 2008), seasonal variations in rainfall (Huang and Chen 2019), large-scale climate variabilities (e.g., El Niño–Southern Oscillation), and future climate change on the island-scale flow response is now being undertaken.
Acknowledgments
This study was funded by the National Science Foundation (NSF) under Grant AGS-1142558 and also partly funded by the Office of Insular Affairs (OIA) Technical Assistance Program under Grant 18AP00090 and the COMET/UCAR Outreach Program under Grant SUBAWD001374. The publication cost is funded by NSF under Grant AGS-1854443. Constructive suggestions by the editors, Drs. David Schultz, Daniel Kirshbaum, and anonymous reviewers have greatly helped to improve the manuscript. We thank Chih-Ying Chen for his assistance, and May Izumi for editing the text.
APPENDIX
Lower Boundary Conditions over Land
a. Soil data and land use
The default soil type and land surface dataset for the Hawaiian region used in the WRF Model does not adequately depict current land use. The soil type data updated by Zhang et al. (2005a, not shown in their paper) from the soil surveyed by the state of Hawaiʻi (Foote et al. 1972; Sato et al. 1973) are used in this study (Fig. A1). Considering the recent housing development of the Second City (Kapolei) (Fig. A2), we adapted the latest C-CAP land use data (NOAA 2019) to match the 24 default categories used by the U.S. Geological Survey (USGS). In the C-CAP data, there are 23 categories. The category conversion is listed in Table A1. After conversion, the 30-m dataset was interpolated into the 1.5-km WRF grid (Fig. A2a). During the interpolation, the dominant vegetation type was decided by the largest grid counts of the converted 30-m dataset within every grid box in our 1.5-km WRF grid. The vegetation fraction was computed as the ratio of the number of converted 30-m grids containing vegetation to the total number of converted 30-m grids within the 1.5-km grid box (Fig. A2b). The coastline of Oahu was also updated with C-CAP high-resolution land cover data. Elevation data are derived from the USGS global 30 arcsec (~900 m) elevation dataset. In some cases, the terrain height at points beyond the coastline are greater than zero due to interpolation. In this case, terrain height is set to zero, and land cover and soil type are indicated as water.
Oahu soil type with 1.5-km grids adapted from the state of Hawaii soil surveys (Foote et al. 1972; Sato et al. 1973).
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Oahu soil type with 1.5-km grids adapted from the state of Hawaii soil surveys (Foote et al. 1972; Sato et al. 1973).
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Oahu soil type with 1.5-km grids adapted from the state of Hawaii soil surveys (Foote et al. 1972; Sato et al. 1973).
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
(a) New land-cover/land-use data for Oahu with 1.5-km grids adapted from the Coastal Change Analysis Program (C-CAP) High-Resolution Land Cover. (b) New vegetation fraction (%) adapted from the C-CAP High-Resolution Land Cover.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
(a) New land-cover/land-use data for Oahu with 1.5-km grids adapted from the Coastal Change Analysis Program (C-CAP) High-Resolution Land Cover. (b) New vegetation fraction (%) adapted from the C-CAP High-Resolution Land Cover.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
(a) New land-cover/land-use data for Oahu with 1.5-km grids adapted from the Coastal Change Analysis Program (C-CAP) High-Resolution Land Cover. (b) New vegetation fraction (%) adapted from the C-CAP High-Resolution Land Cover.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
Conversion table for C-CAP category to USGS category.
b. Soil moisture spinup
In the WRF Model, initial soil moisture and temperature are derived from global model data without considering past precipitation. The CFSR dataset, with its 0.5° grid size, only resolves the Hawaiian Islands with five grids. Oahu does not have its own grid box, resulting in its soil moisture being representative of that found in the grid box covering the neighboring island. Because soil moisture is dependent on past rainfall, the surface parameters (soil temperature and soil moisture) are spun up for one month before initiating the case simulations. Figures A3a and A3b show the initial soil moisture given by the WRF Model and that given after 1 month of spinup. Following Zhang et al. (2005a,b) and Yang et al. (2005), the 24-h simulations of high resolution soil moisture and soil temperature from the model run of the previous day are used to provide updated soil conditions for the next run. The 24-h model simulation (13th to the 36th hour) is used as the simulated diurnal cycle (0300 to 0200 HST the next day) for each day. After the 1-month spinup, initial soil moisture is resolved with more detail, particularly concerning the moist soil over the two mountain ranges and central Oahu (Fig. A3b) due to the occurrence of convective storms on and before 7 June (not shown). The case simulations are begun after running the simulations prior to heavy rainfall events for one month (Fig. A3b).
The soil moisture at 10-cm depth (m3 m−3) for (a) the no soil moisture spinup and (b) after the 1-month spinup at 1400 HST 7 Jun.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
The soil moisture at 10-cm depth (m3 m−3) for (a) the no soil moisture spinup and (b) after the 1-month spinup at 1400 HST 7 Jun.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
The soil moisture at 10-cm depth (m3 m−3) for (a) the no soil moisture spinup and (b) after the 1-month spinup at 1400 HST 7 Jun.
Citation: Monthly Weather Review 148, 10; 10.1175/MWR-D-19-0358.1
REFERENCES
Bingaman, J. B., 2005: Characteristics of the trade wind inversion over Hawaii. M.S. thesis, Dept. of Meteorology, University of Hawaii at Mānoa, 58 pp. [Available from Dept. of Atmos. Sci., 2525 Correa Rd., Honolulu, HI 96822.]
Blumenstock, D. L., and S. Price, 1967: Climates of the United States—Hawaii. Environmental Science Services Administration, Climatography of the United States Rep. 60-51, 27 pp.
Bradshaw, L. S., G. Petrescu, and I. Grenfell, 2003: An initial analysis of relationships between 2- and 10-minute averaged winds at 10, 6, and 1.8 meters: Implications for fire behavior and danger applications. Preprints, Fifth Symp. on Fire and Forest Meteorology, Orlando, FL, Amer. Meteor. Soc., 4.2, http://ams.confex.com/ams/pdfpapers/65901.pdf.
Carbone, R. E., J. W. Wilson, T. D. Keenan, and J. M. Hacker, 2000: Tropical island convection in the absence of significant topography. Part I: Life cycle of diurnally forced convection. Mon. Wea. Rev., 128, 3459–3480, https://doi.org/10.1175/1520-0493(2000)128<3459:TICITA>2.0.CO;2.
Chen, C.-S., Y.-L. Chen, W.-C. Chen, and P.-L. Lin, 2007: Statistics of heavy rainfall occurrences in Taiwan. Wea. Forecasting, 22, 981–1002, https://doi.org/10.1175/WAF1033.1.
Chen, F., and J. Dudhia, 2001: Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129, 569–585, https://doi.org/10.1175/1520-0493(2001)129<0569:CAALSH>2.0.CO;2.
Chen, Y.-L., and J. Feng, 2001: Numerical simulations of airflow and cloud distributions over the windward side of the island of Hawaii. Part I: The effects of trade wind inversion. Mon. Wea. Rev., 129, 1117–1134, https://doi.org/10.1175/1520-0493(2001)129<1117:NSOAAC>2.0.CO;2.
Chen, Y.-L., Y.-J. Chu, C.-S. Chen, C.-C. Tu, J.-H. Teng, and P.-L. Lin, 2018: Analysis and simulations of a heavy rainfall event over northern Taiwan during 11–12 June 2012. Mon. Wea. Rev., 146, 2697–2715, https://doi.org/10.1175/MWR-D-18-0001.1.
Chiao, S., Y.-L. Lin, and M. L. Kaplan, 2004: Numerical study of the orographic forcing of heavy precipitation during MAP IOP-2B. Mon. Wea. Rev., 132, 2184–2203, https://doi.org/10.1175/1520-0493(2004)132<2184:NSOTOF>2.0.CO;2.
Cram, R. S., and H. R. Tatum, 1979: Record torrential rainstorms on the island of Hawaii, January–February 1979. Mon. Wea. Rev., 107, 1653–1662, https://doi.org/10.1175/1520-0493(1979)107<1653:RTROTI>2.0.CO;2.
de Wekker, S., and M. Kossmann, 2015: Convective boundary layer heights over mountainous terrain—A review of concepts. Front. Earth Sci., 3, 77, https://doi.org/10.3389/feart.2015.00077.
Dracup, J. A., E. D. H. Cheng, J. M. Nigg, and T. A. Schroeder, 1991: The New Year’s Eve flood on Oahu, Hawaii, December 31, 1987–January 1, 1988. Natural Disaster Studies, Riley M. Chung, Ed., Vol. 1, National Academy Press, 18–32.
Dudhia, J., 1989: Numerical study of convection observed during the Winter Monsoon Experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46, 3077–3107, https://doi.org/10.1175/1520-0469(1989)046<3077:NSOCOD>2.0.CO;2.
Foote, D. E., E. L. Hill, S. Nakamura, and F. Stephens, 1972: Soil survey of islands of Kauai, Oahu, Maui, Molokaʻi, and Lanai, State of Hawaii. Soil Conservation Service, U.S. Dept. of Agriculture, in cooperation with the University of Hawaii Agricultural Experiment Station, 115 pp. [Available from Natural Resources Conservation Service, U.S. Dept. of Agriculture, Rm. 4-118, 300 Ala Moana Blvd., Honolulu, HI 96850.]
Haraguchi, P., 1977: Forecasting floods in Hawaii (excluding Hawaii island). NOAA Tech. Memo. NWSTM PR-16, 30 pp.
Hartley, T. M., and Y.-L. Chen, 2010: Characteristics of summer trade wind rainfall over Oahu. Wea. Forecasting, 25, 1797–1815, https://doi.org/10.1175/2010WAF2222328.1.
Hitzl, D. E., Y.-L. Chen, and H. V. Nguyen, 2014: Numerical simulations and observations of airflow through the ‘Alenuihāhā channel, Hawaii. Mon. Wea. Rev., 142, 4696–4718, https://doi.org/10.1175/MWR-D-13-00312.1.
Hitzl, D. E., Y.-L. Chen, and F. Hsiao, 2020: Wintertime easterly and southeasterly airflow in the ‘Alenuihāhā Channel. Mon. Wea. Rev., 148, 1337–1362, https://doi.org/10.1175/MWR-D-19-0042.1.
Hong, S.-Y., and J.-O. J. Lim, 2006: The WRF single-moment 6-class microphysics scheme (WSM6). J. Korean Meteor. Soc., 42, 129–151.
Hong, S.-Y., Y. Noh, and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318–2341, https://doi.org/10.1175/MWR3199.1.
Horel, J., and Coauthors, 2002: MesoWest: Cooperative mesonets in the western United States. Bull. Amer. Meteor. Soc., 83, 211–225, https://doi.org/10.1175/1520-0477(2002)083<0211:MCMITW>2.3.CO;2.
Huang, Y.-F., and Y.-L. Chen, 2019: Numerical simulations of seasonal variations of rainfall over the Island of Hawaii. J. Appl. Meteor. Climatol., 58, 1219–1232, https://doi.org/10.1175/JAMC-D-18-0078.1.
Hudson, J. G., 1993: Cloud condensation nuclei near marine cumulus. J. Geophys. Res., 98, 2693–2702, https://doi.org/10.1029/92JD02169.
Janjić, Z. I., 1994: The step-mountain eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev., 122, 927–945, https://doi.org/10.1175/1520-0493(1994)122<0927:TSMECM>2.0.CO;2.
Janjić, Z. I., 2000: Comments on “Development and evaluation of a convection scheme for use in climate models.” J. Atmos. Sci., 57, 3686, https://doi.org/10.1175/1520-0469(2000)057<3686:CODAEO>2.0.CO;2.
Jiménez, P. A., J. Dudhia, J. F. González-Rouco, J. Navarro, J. P. Montávez, and E. Garćia-Bustamante, 2012: A revised scheme for the WRF surface layer formulation. Mon. Wea. Rev., 140, 898–918, https://doi.org/10.1175/MWR-D-11-00056.1.
Jury, M., and S. Chiao, 2013: Leeside boundary layer confluence and afternoon thunderstorms over Mayaguez, Puerto Rico. J. Appl. Meteor. Climatol., 52, 439–454, https://doi.org/10.1175/JAMC-D-11-087.1.
Kerns, B. W. J., Y.-L. Chen, and M.-Y. Chang, 2010: The diurnal cycle of winds, rain, and clouds over Taiwan during the Mei-Yu, summer, and autumn rainfall regimes. Mon. Wea. Rev., 138, 497–516, https://doi.org/10.1175/2009MWR3031.1.
Kodama, K., and G. M. Barnes, 1997: Heavy rain events over the south-facing slopes of Hawaii: Attendant conditions. Wea. Forecasting, 12, 347–367, https://doi.org/10.1175/1520-0434(1997)012<0347:HREOTS>2.0.CO;2.
Li, J., Y.-L. Chen, and W.-C. Lee, 1997: Analysis of a heavy rainfall event during TAMEX. Mon. Wea. Rev., 125, 1060–1082, https://doi.org/10.1175/1520-0493(1997)125<1060:AOAHRE>2.0.CO;2.
Lin, Y.-L., S. Chiao, T.-A. Wang, M. L. Kaplan, and R. P. Weglarz, 2001: Some common ingredients for heavy orographic rainfall. Wea. Forecasting, 16, 633–660, https://doi.org/10.1175/1520-0434(2001)016<0633:SCIFHO>2.0.CO;2.
Medina, S., B. Smull, R. A. Houze Jr., and M. Steiner, 2005: Cross-barrier flow during orographic precipitation events: Results from MAP and IMPROVE. J. Atmos. Sci., 62, 3580–3598, https://doi.org/10.1175/JAS3554.1.
Mlawer, E. J., S. J. Taubman, P. D. Brown, M. J. Iacono, and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102, 16 663–16 682, https://doi.org/10.1029/97JD00237.
NCEI, 2017: Storm events database. Accessed October 2017, https://www.ncdc.noaa.gov/stormevents/.
NOAA, 2019:Coastal Change Analysis Program (C-CAP) Regional Land Cover. NOAA Office for Coastal Management, accessed March 2019, www.coast.noaa.gov/ccapftp.
NWS, 2016: Storm data preparation. National Weather Service Instruction 10-1605, 110 pp., www.nws.noaa.gov/directives/sym/pd01016005curr.pdf.
Nguyen, H. V., Y.-L. Chen, and F. Fujioka, 2010: Numerical simulations of island effects on airflow and weather during the summer over the island of Oahu. Mon. Wea. Rev., 138, 2253–2280, https://doi.org/10.1175/2009MWR3203.1.
Ogura, Y., T. Asai, and K. Dohi, 1985: A case study of a heavy precipitation event along the Baiu front in northern Kyushu, 23 July 1982: Nagasaki heavy rainfall. J. Meteor. Soc. Japan, 63, 883–900, https://doi.org/10.2151/jmsj1965.63.5_883.
Qian, J.-H., 2008: Why precipitation is mostly concentrated over islands in the Maritime Continent. J. Atmos. Sci., 65, 1428–1441, https://doi.org/10.1175/2007JAS2422.1.
Reeve, M. A., and E. W. Kolstad, 2011: The Spitsbergen south cape tip jet. Quart. J. Roy. Meteor. Soc., 137, 1739–1748, https://doi.org/10.1002/qj.876.
Rex, D. F., 1950: Blocking action in the middle troposphere and its effect upon regional climate: II. The climatology of blocking action. Tellus, 2A, 275–301, https://doi.org/10.3402/TELLUSA.V2I4.8603.
Ruppert, J. H., Jr., R. H. Johnson, and A. K. Rowe, 2013: Diurnal circulations and rainfall in Taiwan during SoWMEX/TiMREX (2008). Mon. Wea. Rev., 141, 3851–3872, https://doi.org/10.1175/MWR-D-12-00301.1.
Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 1015–1058, https://doi.org/10.1175/2010BAMS3001.1.
Saha, S., and Coauthors, 2014: The NCEP climate forecast system version 2. J. Climate, 27, 2185–2208, https://doi.org/10.1175/JCLI-D-12-00823.1.
Sato, H. H., W. Ikeda, R. Paeth, R. Smythe, and M. Takehiro Jr., 1973: Soil survey of the island of Hawaii, State of Hawaii. USDA/Soil Conservation Service and University of Hawaii Agricultural Experiment Station, 232 pp. [Available from Natural Resources Conservation Service, U.S. Dept. of Agriculture, Room 4-118, 300 Ala Moana Blvd., Honolulu, HI 96850.]
Schroeder, T. A., 1977: Meteorological analysis of an Oahu flood. Mon. Wea. Rev., 105, 458–468, https://doi.org/10.1175/1520-0493(1977)105<0458:MAOAOF>2.0.CO;2.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., https://doi.org/10.5065/D68S4MVH.
Smith, R. B., P. Schafer, D. J. Kirshbaum, and E. Regina, 2009: Orographic precipitation in the tropics: Experiments in Dominica. J. Atmos. Sci., 66, 1698–1716, https://doi.org/10.1175/2008JAS2920.1.
Smith, R. B., and Coauthors, 2012: Orographic precipitation in the tropics: The Dominica Experiment. Bull. Amer. Meteor. Soc., 93, 1567–1579, https://doi.org/10.1175/BAMS-D-11-00194.1.
Splitt, M. E., and J. D. Horel, 1998: Use of multivariate linear regression for meteorological data analysis and quality assessment in complex terrain. Preprints, 10th Symp. on Meteorological Observations and Instrumentation, Phoenix, AZ, Amer. Meteor. Soc., 359–362.
Tu, C.-C., and Y.-L. Chen, 2011: Favorable conditions for the development of a heavy rainfall event over Oahu during the 2006 wet period. Wea. Forecasting, 26, 280–300, https://doi.org/10.1175/2010WAF2222449.1.
Tu, C.-C., Y.-L. Chen, P.-L. Lin, and Y. Du, 2019: Characteristics of the marine boundary layer jet over the South China Sea during the early summer rainy season of Taiwan. Mon. Wea. Rev., 147, 457–475, https://doi.org/10.1175/MWR-D-18-0230.1.
Tu, C.-C., Y.-L. Chen, P.-L. Lin, and P.-H. Lin, 2020: The relationship between the boundary layer moisture transport from the South China Sea and heavy rainfall over Taiwan. Terr. Atmos. Oceanic Sci., 31, 159–176, https://doi.org/10.3319/TAO.2019.07.01.01.
Wang, C., and D. J. Kirshbaum, 2015: Thermally forced convection over a mountainous tropical island. J. Atmos. Sci., 72, 2484–2506, https://doi.org/10.1175/JAS-D-14-0325.1.
Wang, J.-J., and Y.-L. Chen, 1998: A case study of trade-wind rainbands and their interaction with the island-induced airflow. Mon. Wea. Rev., 126, 409–423, https://doi.org/10.1175/1520-0493(1998)126<0409:ACSOTW>2.0.CO;2.
Wang, S.-T., H. Cheng, C.-H. Shu, and Y.-K. Chao, 1985: The environmental condition for heavy rainfall in Taiwan during May and June. Proc. Conf. on Weather Analysis and Forecasting, Taipei, Taiwan, Central Weather Bureau, 55–88.
Winning, T., Y.-L. Chen, and F. Xie, 2017: Estimation of the marine boundary layer height over the Central North Pacific using GPS radio occultation. Atmos. Res., 183, 362–370, https://doi.org/10.1016/j.atmosres.2016.08.005.
Yang, Y., Y.-L. Chen, and F. M. Fujioka, 2005: Numerical simulations of the island-induced circulations over the island of Hawaii during HaRP. Mon. Wea. Rev., 133, 3693–3713, https://doi.org/10.1175/MWR3053.1.
Yang, Y., Y.-L. Chen, and F. M. Fujioka, 2008: Effects of trade-wind strength and direction on the leeside circulations and rainfall of the Island of Hawaii. Mon. Wea. Rev., 136, 4799–4818, https://doi.org/10.1175/2008MWR2365.1.
Yeh, H.-C., and Y.-L. Chen, 1998: Characteristics of rainfall distributions over Taiwan during the Taiwan Area Mesoscale Experiment (TAMEX). J. Appl. Meteor., 37, 1457–1469, https://doi.org/10.1175/1520-0450(1998)037<1457:CORDOT>2.0.CO;2.
Zhang, Y., Y.-L. Chen, S.-Y. Hong, H.-M. H. Juang, and K. Kodama, 2005a: Validation of the coupled NCEP Meso-scale Spectral Model and an advanced land surface model over the Hawaiian Islands. Part I: Summer trade wind conditions and a heavy rainfall event. Wea. Forecasting, 20, 847–872, https://doi.org/10.1175/WAF891.1.
Zhang, Y., Y.-L. Chen, and K. Kodama, 2005b: Validation of the coupled NCEP mesoscale spectral model and an advanced land surface model over the Hawaiian Islands. Part II: A high wind event. Wea. Forecasting, 20, 873–895, https://doi.org/10.1175/WAF892.1.